Fig 1.
Timeline of data collection and mean perceived stress level for each measurement.
A. Timeline of data collection. The COVID-19 pandemic began to impact everyday life and medical school during Year-1 of cohort 2019 (March 2020). B. Mean perceived stress levels including error bars. Mean (M) stress levels and Standard Errors (SE) are shown for the baseline measurement (M = 25.38, SE = 0.51) and the final measurement (M = 26.00, SE = 0.54) for cohort 2018 and baseline measurement (M = 26.05, SE = 0.79) and the final measurement (M = 30.65, SE = 0.99) for cohort 2019. Differences were assessed within a cohort with a paired t-test. The difference between the final measurement of cohort 2018 and cohort 2019 was assessed in a linear regression model while controlling for baseline measurement, gender, and study performance (Table 2–Model 1). Significant differences between measurements are shown: *** p < .001.
Table 1.
Descriptive statistics of variables included in the analyses.
Table 2.
Results linear regression model with outcome variable stress level (PSS-14) in the final measurement.
Fig 2.
Social support and perceived stress level during the final measurement per cohort.
A. Emotional-informational support. Each dot represents an observation and demonstrates the perceived stress level and the level of emotional-informational support of an individual student. The slope of the regression line represents the effect of emotional-informational support on stress levels. The slope of the regression line in grey for cohort 2018 is equal to -0.23 (non-significant, Table 2–Model 3). The slope of the regression line in black for cohort 2019 is -0.75 (p < .001; p-value is based on a post-hoc analysis), which equals the sum of the regression coefficients for emotional-informational support and emotional-informational support*cohort 2019 = -0.23–0.52 = -0.75 (Table 2–Model 3). B. Club membership. Observed mean perceived stress levels including error bars (M±SE) for students without and with club membership. In the linear regression model, for cohort 2018, no significant effect is present for club membership (coefficient club member = -0.50, non-significant, Table 2–Model 3). For cohort 2019, the linear regression model shows a difference of -3.68 (p < .01; p-value is based on a post-hoc analysis), which is the sum of the regression coefficients for club member and club member*cohort 2019: -0.50–3.18 = -3.68 (Table 2–Model 3).
Fig 3.
Perceived stress level of groups during the final measurement based on club membership and emotional-informational support.
Mean perceived stress level for groups based on club membership and emotional-informational support for cohort 2018 and cohort 2019 including error bars. Emotional-informational support is unchecked if the score is ≤25th percentile of the complete sample (score ≤11). For each cohort separately, the stress levels of groups 2 to 4 were compared to the stress level of group 1 in a post-hoc linear regression analysis (reference group = Ref.). For cohort 2018, groups 2 to 4 did not differ significantly from group 1 regarding stress levels. For cohort 2019, groups 2 to 4 did differ significantly from group 1 regarding stress level: *p < .05, **p < .01, *** p < .001, respectively.